You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@manifoldcf.apache.org by kw...@apache.org on 2014/10/08 17:15:50 UTC
svn commit: r1630136 -
/manifoldcf/branches/CONNECTORS-1067/connectors/livelink/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/livelink/LivelinkConnector.java
Author: kwright
Date: Wed Oct 8 15:15:50 2014
New Revision: 1630136
URL: http://svn.apache.org/r1630136
Log:
Update livelink connector
Modified:
manifoldcf/branches/CONNECTORS-1067/connectors/livelink/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/livelink/LivelinkConnector.java
Modified: manifoldcf/branches/CONNECTORS-1067/connectors/livelink/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/livelink/LivelinkConnector.java
URL: http://svn.apache.org/viewvc/manifoldcf/branches/CONNECTORS-1067/connectors/livelink/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/livelink/LivelinkConnector.java?rev=1630136&r1=1630135&r2=1630136&view=diff
==============================================================================
--- manifoldcf/branches/CONNECTORS-1067/connectors/livelink/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/livelink/LivelinkConnector.java (original)
+++ manifoldcf/branches/CONNECTORS-1067/connectors/livelink/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/livelink/LivelinkConnector.java Wed Oct 8 15:15:50 2014
@@ -4185,466 +4185,485 @@ public class LivelinkConnector extends o
try
{
// Check URL first
- if (activities.checkURLIndexable(viewHttpAddress))
+ if (!activities.checkURLIndexable(viewHttpAddress))
{
+ // Document not ingestable due to URL
+ resultDescription = "URL ("+viewHttpAddress+") was rejected by output connector";
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("Livelink: Excluding document "+documentIdentifier+" because its URL ("+viewHttpAddress+") was rejected by output connector");
+ resultCode = "URLEXCLUSION";
+ activities.noDocument(documentIdentifier,version);
+ return;
+ }
+
+ // Add general metadata
+ ObjectInformation objInfo = llc.getObjectInformation(vol,objID);
+ VersionInformation versInfo = llc.getVersionInformation(vol,objID,0);
+ if (!objInfo.exists())
+ {
+ resultCode = "OBJECTNOTFOUND";
+ Logging.connectors.debug("Livelink: No object "+contextMsg+": not ingesting");
+ return;
+ }
+ if (!versInfo.exists())
+ {
+ resultCode = "VERSIONNOTFOUND";
+ Logging.connectors.debug("Livelink: No version data "+contextMsg+": not ingesting");
+ return;
+ }
- // Add general metadata
- ObjectInformation objInfo = llc.getObjectInformation(vol,objID);
- VersionInformation versInfo = llc.getVersionInformation(vol,objID,0);
- if (!objInfo.exists())
- {
- resultCode = "OBJECTNOTFOUND";
- Logging.connectors.debug("Livelink: No object "+contextMsg+": not ingesting");
- return;
- }
- if (!versInfo.exists())
- {
- resultCode = "VERSIONNOTFOUND";
- Logging.connectors.debug("Livelink: No version data "+contextMsg+": not ingesting");
- return;
- }
+ String mimeType = versInfo.getMimeType();
+ if (!activities.checkMimeTypeIndexable(mimeType))
+ {
+ // Document not indexable because of its mime type
+ resultDescription = "Mime type ("+mimeType+") was rejected by output connector";
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("Livelink: Excluding document "+documentIdentifier+" because its mime type ("+mimeType+") was rejected by output connector");
+ resultCode = "MIMETYPEEXCLUSION";
+ activities.noDocument(documentIdentifier,version);
+ return;
+ }
+
+ Long dataSize = versInfo.getDataSize();
+ if (dataSize == null)
+ {
+ // Document had no length
+ resultDescription = "Document had no length";
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("Livelink: Excluding document "+documentIdentifier+" because it had no length");
+ resultCode = "DOCUMENTNOLENGTH";
+ activities.noDocument(documentIdentifier,version);
+ return;
+ }
+
+ if (!activities.checkLengthIndexable(dataSize.longValue()))
+ {
+ // Document not indexable because of its length
+ resultDescription = "Document length ("+dataSize+") was rejected by output connector";
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("Livelink: Excluding document "+documentIdentifier+" because its length ("+dataSize+") was rejected by output connector");
+ resultCode = "DOCUMENTTOOLONG";
+ activities.noDocument(documentIdentifier,version);
+ return;
+ }
- String mimeType = versInfo.getMimeType();
- if (activities.checkMimeTypeIndexable(mimeType))
- {
- Long dataSize = versInfo.getDataSize();
- if (dataSize != null && activities.checkLengthIndexable(dataSize.longValue()))
- {
- String fileName = versInfo.getFileName();
- Date creationDate = objInfo.getCreationDate();
- Date modifyDate = versInfo.getModifyDate();
- Integer parentID = objInfo.getParentId();
- RepositoryDocument rd = new RepositoryDocument();
+ Date modifyDate = versInfo.getModifyDate();
+ if (!activities.checkDateIndexable(modifyDate))
+ {
+ // Document not indexable because of its date
+ resultDescription = "Document date ("+modifyDate+") was rejected by output connector";
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("Livelink: Excluding document "+documentIdentifier+" because its date ("+modifyDate+") was rejected by output connector");
+ resultCode = "DOCUMENTBADDATE";
+ activities.noDocument(documentIdentifier,version);
+ return;
+ }
+
+ String fileName = versInfo.getFileName();
+ Date creationDate = objInfo.getCreationDate();
+ Integer parentID = objInfo.getParentId();
+
+
+ RepositoryDocument rd = new RepositoryDocument();
+ // Add general data we need for the output connector
+ if (mimeType != null)
+ rd.setMimeType(mimeType);
+ if (fileName != null)
+ rd.setFileName(fileName);
+ if (creationDate != null)
+ rd.setCreatedDate(creationDate);
+ if (modifyDate != null)
+ rd.setModifiedDate(modifyDate);
- // Add general data we need for the output connector
- if (mimeType != null)
- rd.setMimeType(mimeType);
- if (fileName != null)
- rd.setFileName(fileName);
- if (creationDate != null)
- rd.setCreatedDate(creationDate);
- if (modifyDate != null)
- rd.setModifiedDate(modifyDate);
-
- rd.addField(GENERAL_NAME_FIELD,objInfo.getName());
- rd.addField(GENERAL_DESCRIPTION_FIELD,objInfo.getComments());
- if (creationDate != null)
- rd.addField(GENERAL_CREATIONDATE_FIELD,new Long(creationDate.getTime()).toString());
- if (modifyDate != null)
- rd.addField(GENERAL_MODIFYDATE_FIELD,new Long(modifyDate.getTime()).toString());
- if (parentID != null)
- rd.addField(GENERAL_PARENTID,parentID.toString());
- UserInformation owner = llc.getUserInformation(objInfo.getOwnerId().intValue());
- UserInformation creator = llc.getUserInformation(objInfo.getCreatorId().intValue());
- UserInformation modifier = llc.getUserInformation(versInfo.getOwnerId().intValue());
- if (owner != null)
- rd.addField(GENERAL_OWNER,owner.getName());
- if (creator != null)
- rd.addField(GENERAL_CREATOR,creator.getName());
- if (modifier != null)
- rd.addField(GENERAL_MODIFIER,modifier.getName());
-
- // Iterate over the metadata items. These are organized by category
- // for speed of lookup.
-
- Iterator<MetadataItem> catIter = desc.getItems(categoryPaths);
- while (catIter.hasNext())
+ rd.addField(GENERAL_NAME_FIELD,objInfo.getName());
+ rd.addField(GENERAL_DESCRIPTION_FIELD,objInfo.getComments());
+ if (creationDate != null)
+ rd.addField(GENERAL_CREATIONDATE_FIELD,new Long(creationDate.getTime()).toString());
+ if (modifyDate != null)
+ rd.addField(GENERAL_MODIFYDATE_FIELD,new Long(modifyDate.getTime()).toString());
+ if (parentID != null)
+ rd.addField(GENERAL_PARENTID,parentID.toString());
+ UserInformation owner = llc.getUserInformation(objInfo.getOwnerId().intValue());
+ UserInformation creator = llc.getUserInformation(objInfo.getCreatorId().intValue());
+ UserInformation modifier = llc.getUserInformation(versInfo.getOwnerId().intValue());
+ if (owner != null)
+ rd.addField(GENERAL_OWNER,owner.getName());
+ if (creator != null)
+ rd.addField(GENERAL_CREATOR,creator.getName());
+ if (modifier != null)
+ rd.addField(GENERAL_MODIFIER,modifier.getName());
+
+ // Iterate over the metadata items. These are organized by category
+ // for speed of lookup.
+
+ Iterator<MetadataItem> catIter = desc.getItems(categoryPaths);
+ while (catIter.hasNext())
+ {
+ MetadataItem item = catIter.next();
+ MetadataPathItem pathItem = item.getPathItem();
+ if (pathItem != null)
+ {
+ int catID = pathItem.getCatID();
+ // grab the associated catversion
+ LLValue catVersion = getCatVersion(objID,catID);
+ if (catVersion != null)
+ {
+ // Go through attributes now
+ Iterator<String> attrIter = item.getAttributeNames();
+ while (attrIter.hasNext())
{
- MetadataItem item = catIter.next();
- MetadataPathItem pathItem = item.getPathItem();
- if (pathItem != null)
- {
- int catID = pathItem.getCatID();
- // grab the associated catversion
- LLValue catVersion = getCatVersion(objID,catID);
- if (catVersion != null)
- {
- // Go through attributes now
- Iterator<String> attrIter = item.getAttributeNames();
- while (attrIter.hasNext())
- {
- String attrName = attrIter.next();
- // Create a unique metadata name
- String metadataName = pathItem.getCatName()+":"+attrName;
- // Fetch the metadata and stuff it into the RepositoryData structure
- String[] metadataValue = getAttributeValue(catVersion,attrName);
- if (metadataValue != null)
- rd.addField(metadataName,metadataValue);
- else
- Logging.connectors.warn("Livelink: Metadata attribute '"+metadataName+"' does not seem to exist; please correct the job");
- }
- }
-
- }
+ String attrName = attrIter.next();
+ // Create a unique metadata name
+ String metadataName = pathItem.getCatName()+":"+attrName;
+ // Fetch the metadata and stuff it into the RepositoryData structure
+ String[] metadataValue = getAttributeValue(catVersion,attrName);
+ if (metadataValue != null)
+ rd.addField(metadataName,metadataValue);
+ else
+ Logging.connectors.warn("Livelink: Metadata attribute '"+metadataName+"' does not seem to exist; please correct the job");
}
+ }
+
+ }
+ }
- if (actualAcls != null && denyAcls != null)
- rd.setSecurity(RepositoryDocument.SECURITY_TYPE_DOCUMENT,actualAcls,denyAcls);
+ if (actualAcls != null && denyAcls != null)
+ rd.setSecurity(RepositoryDocument.SECURITY_TYPE_DOCUMENT,actualAcls,denyAcls);
- // Add the path metadata item into the mix, if enabled
- String pathAttributeName = sDesc.getPathAttributeName();
- if (pathAttributeName != null && pathAttributeName.length() > 0)
- {
- String pathString = sDesc.getPathAttributeValue(documentIdentifier);
- if (pathString != null)
- {
- if (Logging.connectors.isDebugEnabled())
- Logging.connectors.debug("Livelink: Path attribute name is '"+pathAttributeName+"'"+contextMsg+", value is '"+pathString+"'");
- rd.addField(pathAttributeName,pathString);
- }
- }
+ // Add the path metadata item into the mix, if enabled
+ String pathAttributeName = sDesc.getPathAttributeName();
+ if (pathAttributeName != null && pathAttributeName.length() > 0)
+ {
+ String pathString = sDesc.getPathAttributeValue(documentIdentifier);
+ if (pathString != null)
+ {
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("Livelink: Path attribute name is '"+pathAttributeName+"'"+contextMsg+", value is '"+pathString+"'");
+ rd.addField(pathAttributeName,pathString);
+ }
+ }
- if (ingestProtocol != null)
- {
- // Use HTTP to fetch document!
- String ingestHttpAddress = convertToIngestURI(documentIdentifier);
- if (ingestHttpAddress != null)
- {
+ if (ingestProtocol != null)
+ {
+ // Use HTTP to fetch document!
+ String ingestHttpAddress = convertToIngestURI(documentIdentifier);
+ if (ingestHttpAddress != null)
+ {
- // Set up connection
- HttpClient client = getInitializedClient(contextMsg);
+ // Set up connection
+ HttpClient client = getInitializedClient(contextMsg);
- long currentTime;
+ long currentTime;
- if (Logging.connectors.isInfoEnabled())
- Logging.connectors.info("Livelink: " + ingestHttpAddress);
+ if (Logging.connectors.isInfoEnabled())
+ Logging.connectors.info("Livelink: " + ingestHttpAddress);
- HttpGet method = new HttpGet(getHost().toURI() + ingestHttpAddress);
- method.setHeader(new BasicHeader("Accept","*/*"));
+ HttpGet method = new HttpGet(getHost().toURI() + ingestHttpAddress);
+ method.setHeader(new BasicHeader("Accept","*/*"));
- ExecuteMethodThread methodThread = new ExecuteMethodThread(client,method);
- methodThread.start();
+ ExecuteMethodThread methodThread = new ExecuteMethodThread(client,method);
+ methodThread.start();
+ try
+ {
+
+ int statusCode = methodThread.getResponseCode();
+ switch (statusCode)
+ {
+ case 500:
+ case 502:
+ Logging.connectors.warn("Livelink: Service interruption during fetch "+contextMsg+" with Livelink HTTP Server, retrying...");
+ throw new ServiceInterruption("Service interruption during fetch",new ManifoldCFException(Integer.toString(statusCode)+" error while fetching"),System.currentTimeMillis()+60000L,
+ System.currentTimeMillis()+600000L,-1,true);
+
+ case HttpStatus.SC_UNAUTHORIZED:
+ Logging.connectors.warn("Livelink: Document fetch unauthorized for "+ingestHttpAddress+" ("+contextMsg+")");
+ // Since we logged in, we should fail here if the ingestion user doesn't have access to the
+ // the document, but if we do, don't fail hard.
+ resultCode = "UNAUTHORIZED";
+ activities.noDocument(documentIdentifier,version);
+ return;
+
+ case HttpStatus.SC_OK:
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("Livelink: Created http document connection to Livelink "+contextMsg);
+ // A non-existent content length will cause a value of -1 to be returned. This seems to indicate that the session login did not work right.
+ if (methodThread.getResponseContentLength() >= 0)
+ {
try
{
-
- int statusCode = methodThread.getResponseCode();
- switch (statusCode)
+ InputStream is = methodThread.getSafeInputStream();
+ try
{
- case 500:
- case 502:
- Logging.connectors.warn("Livelink: Service interruption during fetch "+contextMsg+" with Livelink HTTP Server, retrying...");
- throw new ServiceInterruption("Service interruption during fetch",new ManifoldCFException(Integer.toString(statusCode)+" error while fetching"),System.currentTimeMillis()+60000L,
- System.currentTimeMillis()+600000L,-1,true);
-
- case HttpStatus.SC_UNAUTHORIZED:
- Logging.connectors.warn("Livelink: Document fetch unauthorized for "+ingestHttpAddress+" ("+contextMsg+")");
- // Since we logged in, we should fail here if the ingestion user doesn't have access to the
- // the document, but if we do, don't fail hard.
- resultCode = "UNAUTHORIZED";
- activities.noDocument(documentIdentifier,version);
- return;
-
- case HttpStatus.SC_OK:
- if (Logging.connectors.isDebugEnabled())
- Logging.connectors.debug("Livelink: Created http document connection to Livelink "+contextMsg);
- // A non-existent content length will cause a value of -1 to be returned. This seems to indicate that the session login did not work right.
- if (methodThread.getResponseContentLength() >= 0)
- {
- try
- {
- InputStream is = methodThread.getSafeInputStream();
- try
- {
- rd.setBinary(is,dataSize);
+ rd.setBinary(is,dataSize);
- activities.ingestDocumentWithException(documentIdentifier,version,viewHttpAddress,rd);
-
- if (Logging.connectors.isDebugEnabled())
- Logging.connectors.debug("Livelink: Ingesting done "+contextMsg);
+ activities.ingestDocumentWithException(documentIdentifier,version,viewHttpAddress,rd);
+
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("Livelink: Ingesting done "+contextMsg);
- }
- finally
- {
- // Close stream via thread, since otherwise this can hang
- is.close();
- }
- }
- catch (java.net.SocketTimeoutException e)
- {
- resultCode = "DATATIMEOUT";
- resultDescription = e.getMessage();
- currentTime = System.currentTimeMillis();
- Logging.connectors.warn("Livelink: Livelink socket timed out ingesting from the Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
- throw new ServiceInterruption("Socket timed out: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,false);
- }
- catch (java.net.SocketException e)
- {
- resultCode = "DATASOCKETERROR";
- resultDescription = e.getMessage();
- currentTime = System.currentTimeMillis();
- Logging.connectors.warn("Livelink: Livelink socket error ingesting from the Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
- throw new ServiceInterruption("Socket error: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,false);
- }
- catch (javax.net.ssl.SSLHandshakeException e)
- {
- resultCode = "DATASSLHANDSHAKEERROR";
- resultDescription = e.getMessage();
- currentTime = System.currentTimeMillis();
- Logging.connectors.warn("Livelink: SSL handshake failed authenticating "+contextMsg+": "+e.getMessage(),e);
- throw new ServiceInterruption("SSL handshake error: "+e.getMessage(),e,currentTime+60000L,currentTime+300000L,-1,true);
- }
- catch (ConnectTimeoutException e)
- {
- resultCode = "CONNECTTIMEOUT";
- resultDescription = e.getMessage();
- currentTime = System.currentTimeMillis();
- Logging.connectors.warn("Livelink: Livelink socket timed out connecting to the Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
- throw new ServiceInterruption("Connect timed out: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,false);
- }
- catch (InterruptedException e)
- {
- wasInterrupted = true;
- throw new ManifoldCFException("Interrupted: "+e.getMessage(),e,ManifoldCFException.INTERRUPTED);
- }
- catch (InterruptedIOException e)
- {
- wasInterrupted = true;
- throw new ManifoldCFException("Interrupted: "+e.getMessage(),e,ManifoldCFException.INTERRUPTED);
- }
- catch (HttpException e)
- {
- resultCode = "HTTPEXCEPTION";
- resultDescription = e.getMessage();
- // Treat unknown error ingesting data as a transient condition
- currentTime = System.currentTimeMillis();
- Logging.connectors.warn("Livelink: HTTP exception ingesting "+contextMsg+": "+e.getMessage(),e);
- throw new ServiceInterruption("HTTP exception ingesting "+contextMsg+": "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,false);
- }
- catch (IOException e)
- {
- resultCode = "DATAEXCEPTION";
- resultDescription = e.getMessage();
- // Treat unknown error ingesting data as a transient condition
- currentTime = System.currentTimeMillis();
- Logging.connectors.warn("Livelink: IO exception ingesting "+contextMsg+": "+e.getMessage(),e);
- throw new ServiceInterruption("IO exception ingesting "+contextMsg+": "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,false);
- }
- readSize = dataSize;
- }
- else
- {
- resultCode = "SESSIONLOGINFAILED";
- activities.noDocument(documentIdentifier,version);
- }
- break;
- case HttpStatus.SC_BAD_REQUEST:
- case HttpStatus.SC_USE_PROXY:
- case HttpStatus.SC_GONE:
- resultCode = "ERROR "+Integer.toString(statusCode);
- throw new ManifoldCFException("Unrecoverable request failure; error = "+Integer.toString(statusCode));
- default:
- resultCode = "UNKNOWN";
- Logging.connectors.warn("Livelink: Attempt to retrieve document from '"+ingestHttpAddress+"' received a response of "+Integer.toString(statusCode)+"; retrying in one minute");
- currentTime = System.currentTimeMillis();
- throw new ServiceInterruption("Fetch failed; retrying in 1 minute",new ManifoldCFException("Fetch failed with unknown code "+Integer.toString(statusCode)),
- currentTime+60000L,currentTime+600000L,-1,true);
}
- }
- catch (InterruptedException e)
- {
- // Drop the connection on the floor
- methodThread.interrupt();
- methodThread = null;
- throw new ManifoldCFException("Interrupted: "+e.getMessage(),e,ManifoldCFException.INTERRUPTED);
+ finally
+ {
+ // Close stream via thread, since otherwise this can hang
+ is.close();
+ }
}
catch (java.net.SocketTimeoutException e)
{
- Logging.connectors.warn("Livelink: Socket timed out reading from the Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
- resultCode = "TIMEOUT";
+ resultCode = "DATATIMEOUT";
resultDescription = e.getMessage();
currentTime = System.currentTimeMillis();
- throw new ServiceInterruption("Socket timed out: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,true);
+ Logging.connectors.warn("Livelink: Livelink socket timed out ingesting from the Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
+ throw new ServiceInterruption("Socket timed out: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,false);
}
catch (java.net.SocketException e)
{
- Logging.connectors.warn("Livelink: Socket error reading from Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
- resultCode = "SOCKETERROR";
+ resultCode = "DATASOCKETERROR";
resultDescription = e.getMessage();
currentTime = System.currentTimeMillis();
- throw new ServiceInterruption("Socket error: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,true);
+ Logging.connectors.warn("Livelink: Livelink socket error ingesting from the Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
+ throw new ServiceInterruption("Socket error: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,false);
}
catch (javax.net.ssl.SSLHandshakeException e)
{
- currentTime = System.currentTimeMillis();
- Logging.connectors.warn("Livelink: SSL handshake failed "+contextMsg+": "+e.getMessage(),e);
- resultCode = "SSLHANDSHAKEERROR";
+ resultCode = "DATASSLHANDSHAKEERROR";
resultDescription = e.getMessage();
+ currentTime = System.currentTimeMillis();
+ Logging.connectors.warn("Livelink: SSL handshake failed authenticating "+contextMsg+": "+e.getMessage(),e);
throw new ServiceInterruption("SSL handshake error: "+e.getMessage(),e,currentTime+60000L,currentTime+300000L,-1,true);
}
catch (ConnectTimeoutException e)
{
- Logging.connectors.warn("Livelink: Connect timed out reading from the Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
resultCode = "CONNECTTIMEOUT";
resultDescription = e.getMessage();
currentTime = System.currentTimeMillis();
- throw new ServiceInterruption("Connect timed out: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,true);
+ Logging.connectors.warn("Livelink: Livelink socket timed out connecting to the Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
+ throw new ServiceInterruption("Connect timed out: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,false);
+ }
+ catch (InterruptedException e)
+ {
+ wasInterrupted = true;
+ throw new ManifoldCFException("Interrupted: "+e.getMessage(),e,ManifoldCFException.INTERRUPTED);
}
catch (InterruptedIOException e)
{
- methodThread.interrupt();
+ wasInterrupted = true;
throw new ManifoldCFException("Interrupted: "+e.getMessage(),e,ManifoldCFException.INTERRUPTED);
}
catch (HttpException e)
{
- resultCode = "EXCEPTION";
+ resultCode = "HTTPEXCEPTION";
resultDescription = e.getMessage();
- throw new ManifoldCFException("Exception getting response "+contextMsg+": "+e.getMessage(), e);
+ // Treat unknown error ingesting data as a transient condition
+ currentTime = System.currentTimeMillis();
+ Logging.connectors.warn("Livelink: HTTP exception ingesting "+contextMsg+": "+e.getMessage(),e);
+ throw new ServiceInterruption("HTTP exception ingesting "+contextMsg+": "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,false);
}
catch (IOException e)
{
- resultCode = "EXCEPTION";
+ resultCode = "DATAEXCEPTION";
resultDescription = e.getMessage();
- throw new ManifoldCFException("Exception getting response "+contextMsg+": "+e.getMessage(), e);
- }
- finally
- {
- if (methodThread != null)
- {
- methodThread.abort();
- if (!wasInterrupted)
- {
- try
- {
- methodThread.finishUp();
- }
- catch (InterruptedException e)
- {
- wasInterrupted = true;
- throw new ManifoldCFException(e.getMessage(),e,ManifoldCFException.INTERRUPTED);
- }
- }
- }
+ // Treat unknown error ingesting data as a transient condition
+ currentTime = System.currentTimeMillis();
+ Logging.connectors.warn("Livelink: IO exception ingesting "+contextMsg+": "+e.getMessage(),e);
+ throw new ServiceInterruption("IO exception ingesting "+contextMsg+": "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,false);
}
+ readSize = dataSize;
}
else
{
- if (Logging.connectors.isDebugEnabled())
- Logging.connectors.debug("Livelink: No fetch URI "+contextMsg+" - not ingesting");
- resultCode = "NOURI";
- return;
+ resultCode = "SESSIONLOGINFAILED";
+ activities.noDocument(documentIdentifier,version);
}
+ break;
+ case HttpStatus.SC_BAD_REQUEST:
+ case HttpStatus.SC_USE_PROXY:
+ case HttpStatus.SC_GONE:
+ resultCode = "ERROR "+Integer.toString(statusCode);
+ throw new ManifoldCFException("Unrecoverable request failure; error = "+Integer.toString(statusCode));
+ default:
+ resultCode = "UNKNOWN";
+ Logging.connectors.warn("Livelink: Attempt to retrieve document from '"+ingestHttpAddress+"' received a response of "+Integer.toString(statusCode)+"; retrying in one minute");
+ currentTime = System.currentTimeMillis();
+ throw new ServiceInterruption("Fetch failed; retrying in 1 minute",new ManifoldCFException("Fetch failed with unknown code "+Integer.toString(statusCode)),
+ currentTime+60000L,currentTime+600000L,-1,true);
}
- else
+ }
+ catch (InterruptedException e)
+ {
+ // Drop the connection on the floor
+ methodThread.interrupt();
+ methodThread = null;
+ throw new ManifoldCFException("Interrupted: "+e.getMessage(),e,ManifoldCFException.INTERRUPTED);
+ }
+ catch (java.net.SocketTimeoutException e)
+ {
+ Logging.connectors.warn("Livelink: Socket timed out reading from the Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
+ resultCode = "TIMEOUT";
+ resultDescription = e.getMessage();
+ currentTime = System.currentTimeMillis();
+ throw new ServiceInterruption("Socket timed out: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,true);
+ }
+ catch (java.net.SocketException e)
+ {
+ Logging.connectors.warn("Livelink: Socket error reading from Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
+ resultCode = "SOCKETERROR";
+ resultDescription = e.getMessage();
+ currentTime = System.currentTimeMillis();
+ throw new ServiceInterruption("Socket error: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,true);
+ }
+ catch (javax.net.ssl.SSLHandshakeException e)
+ {
+ currentTime = System.currentTimeMillis();
+ Logging.connectors.warn("Livelink: SSL handshake failed "+contextMsg+": "+e.getMessage(),e);
+ resultCode = "SSLHANDSHAKEERROR";
+ resultDescription = e.getMessage();
+ throw new ServiceInterruption("SSL handshake error: "+e.getMessage(),e,currentTime+60000L,currentTime+300000L,-1,true);
+ }
+ catch (ConnectTimeoutException e)
+ {
+ Logging.connectors.warn("Livelink: Connect timed out reading from the Livelink HTTP Server "+contextMsg+": "+e.getMessage(), e);
+ resultCode = "CONNECTTIMEOUT";
+ resultDescription = e.getMessage();
+ currentTime = System.currentTimeMillis();
+ throw new ServiceInterruption("Connect timed out: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,true);
+ }
+ catch (InterruptedIOException e)
+ {
+ methodThread.interrupt();
+ throw new ManifoldCFException("Interrupted: "+e.getMessage(),e,ManifoldCFException.INTERRUPTED);
+ }
+ catch (HttpException e)
+ {
+ resultCode = "EXCEPTION";
+ resultDescription = e.getMessage();
+ throw new ManifoldCFException("Exception getting response "+contextMsg+": "+e.getMessage(), e);
+ }
+ catch (IOException e)
+ {
+ resultCode = "EXCEPTION";
+ resultDescription = e.getMessage();
+ throw new ManifoldCFException("Exception getting response "+contextMsg+": "+e.getMessage(), e);
+ }
+ finally
+ {
+ if (methodThread != null)
{
- // Use FetchVersion instead
- long currentTime;
-
- // Fire up the document reading thread
- DocumentReadingThread t = new DocumentReadingThread(vol,objID,0);
- try
+ methodThread.abort();
+ if (!wasInterrupted)
{
- t.start();
try
{
- InputStream is = t.getSafeInputStream();
- try
- {
- // Can only index while background thread is running!
- rd.setBinary(is, dataSize);
- activities.ingestDocumentWithException(documentIdentifier, version, viewHttpAddress, rd);
- }
- finally
- {
- is.close();
- }
- }
- catch (ManifoldCFException e)
- {
- if (e.getErrorCode() == ManifoldCFException.INTERRUPTED)
- wasInterrupted = true;
- throw e;
+ methodThread.finishUp();
}
- catch (java.net.SocketTimeoutException e)
- {
- throw e;
- }
- catch (InterruptedIOException e)
+ catch (InterruptedException e)
{
wasInterrupted = true;
- throw e;
- }
- finally
- {
- if (!wasInterrupted)
- t.finishUp();
- }
-
- // No errors. Record the fact that we made it.
- resultCode = "OK";
- readSize = dataSize;
- }
- catch (InterruptedException e)
- {
- t.interrupt();
- throw new ManifoldCFException("Interrupted: " + e.getMessage(), e,
- ManifoldCFException.INTERRUPTED);
- }
- catch (ConnectTimeoutException e)
- {
- Logging.connectors.warn("Livelink: Connect timed out "+contextMsg+": "+e.getMessage(), e);
- resultCode = "CONNECTTIMEOUT";
- resultDescription = e.getMessage();
- currentTime = System.currentTimeMillis();
- throw new ServiceInterruption("Connect timed out: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,true);
- }
- catch (InterruptedIOException e)
- {
- t.interrupt();
- throw new ManifoldCFException("Interrupted: "+e.getMessage(),e,ManifoldCFException.INTERRUPTED);
- }
- catch (IOException e)
- {
- resultCode = "EXCEPTION";
- resultDescription = e.getMessage();
- throw new ManifoldCFException("Exception getting response "+contextMsg+": "+e.getMessage(), e);
- }
- catch (ManifoldCFException e)
- {
- if (e.getErrorCode() != ManifoldCFException.INTERRUPTED)
- {
- resultCode = "EXCEPTION";
- resultDescription = e.getMessage();
+ throw new ManifoldCFException(e.getMessage(),e,ManifoldCFException.INTERRUPTED);
}
- throw e;
- }
- catch (RuntimeException e)
- {
- resultCode = "EXCEPTION";
- resultDescription = e.getMessage();
- handleLivelinkRuntimeException(e,0,true);
}
}
}
- else
- {
- // Document not indexable because of its length
- resultDescription = "Document length ("+dataSize+") was rejected by output connector";
- if (Logging.connectors.isDebugEnabled())
- Logging.connectors.debug("Livelink: Excluding document "+documentIdentifier+" because its length ("+dataSize+") was rejected by output connector");
- resultCode = "DOCUMENTTOOLONG";
- activities.noDocument(documentIdentifier,version);
- }
}
else
{
- // Document not indexable because of its mime type
- resultDescription = "Mime type ("+mimeType+") was rejected by output connector";
if (Logging.connectors.isDebugEnabled())
- Logging.connectors.debug("Livelink: Excluding document "+documentIdentifier+" because its mime type ("+mimeType+") was rejected by output connector");
- resultCode = "MIMETYPEEXCLUSION";
- activities.noDocument(documentIdentifier,version);
+ Logging.connectors.debug("Livelink: No fetch URI "+contextMsg+" - not ingesting");
+ resultCode = "NOURI";
+ return;
}
}
else
{
- // Document not ingestable due to URL
- resultDescription = "URL ("+viewHttpAddress+") was rejected by output connector";
- if (Logging.connectors.isDebugEnabled())
- Logging.connectors.debug("Livelink: Excluding document "+documentIdentifier+" because its URL ("+viewHttpAddress+") was rejected by output connector");
- resultCode = "URLEXCLUSION";
- activities.noDocument(documentIdentifier,version);
+ // Use FetchVersion instead
+ long currentTime;
+
+ // Fire up the document reading thread
+ DocumentReadingThread t = new DocumentReadingThread(vol,objID,0);
+ try
+ {
+ t.start();
+ try
+ {
+ InputStream is = t.getSafeInputStream();
+ try
+ {
+ // Can only index while background thread is running!
+ rd.setBinary(is, dataSize);
+ activities.ingestDocumentWithException(documentIdentifier, version, viewHttpAddress, rd);
+ }
+ finally
+ {
+ is.close();
+ }
+ }
+ catch (ManifoldCFException e)
+ {
+ if (e.getErrorCode() == ManifoldCFException.INTERRUPTED)
+ wasInterrupted = true;
+ throw e;
+ }
+ catch (java.net.SocketTimeoutException e)
+ {
+ throw e;
+ }
+ catch (InterruptedIOException e)
+ {
+ wasInterrupted = true;
+ throw e;
+ }
+ finally
+ {
+ if (!wasInterrupted)
+ t.finishUp();
+ }
+
+ // No errors. Record the fact that we made it.
+ resultCode = "OK";
+ readSize = dataSize;
+ }
+ catch (InterruptedException e)
+ {
+ t.interrupt();
+ throw new ManifoldCFException("Interrupted: " + e.getMessage(), e,
+ ManifoldCFException.INTERRUPTED);
+ }
+ catch (ConnectTimeoutException e)
+ {
+ Logging.connectors.warn("Livelink: Connect timed out "+contextMsg+": "+e.getMessage(), e);
+ resultCode = "CONNECTTIMEOUT";
+ resultDescription = e.getMessage();
+ currentTime = System.currentTimeMillis();
+ throw new ServiceInterruption("Connect timed out: "+e.getMessage(),e,currentTime+300000L,currentTime+6*3600000L,-1,true);
+ }
+ catch (InterruptedIOException e)
+ {
+ t.interrupt();
+ throw new ManifoldCFException("Interrupted: "+e.getMessage(),e,ManifoldCFException.INTERRUPTED);
+ }
+ catch (IOException e)
+ {
+ resultCode = "EXCEPTION";
+ resultDescription = e.getMessage();
+ throw new ManifoldCFException("Exception getting response "+contextMsg+": "+e.getMessage(), e);
+ }
+ catch (ManifoldCFException e)
+ {
+ if (e.getErrorCode() != ManifoldCFException.INTERRUPTED)
+ {
+ resultCode = "EXCEPTION";
+ resultDescription = e.getMessage();
+ }
+ throw e;
+ }
+ catch (RuntimeException e)
+ {
+ resultCode = "EXCEPTION";
+ resultDescription = e.getMessage();
+ handleLivelinkRuntimeException(e,0,true);
+ }
}
}
finally